New here? Start with the free 60-second AI roadmap for software owners.
Every business for sale has two profit figures. The one in the accounts, and the one in the deck.
The gap between them is where most of the negotiation actually happens, and it is filled with add-backs: costs the seller argues a new owner would not incur. Some are entirely fair. Some are optimistic. A few are fiction that everyone has agreed to treat politely.
The number you need is neither. It is the cash the business will genuinely produce for whoever owns it next.
The most common adjustment, and the one most often done badly in both directions.
A founder-run business rarely pays its founder a market wage. Sometimes far below, where the owner takes distributions and a nominal salary. Sometimes far above, because the company is a tax-efficient way to pay themselves.
Either way the adjustment is not "add back what the founder took." It is: what would it cost to employ someone to do what the founder actually does?
That question is harder than it looks, because the honest answer is usually more than one person. A founder who sells, supports the top accounts, approves every hire, and still reviews code is three roles. Replacing them at market rate might cost considerably more than they were paying themselves, which turns a proposed add-back into a genuine cost increase.
This single adjustment moves the number more than any other in most small software businesses, and it is the one worth spending a week on rather than an afternoon.
The second category is costs described as exceptional.
Legal fees for a dispute. A failed hire. A migration. A write-off. Each is genuinely non-recurring in the narrow sense that this particular one will not repeat.
But look back four years. If there is a settlement in three of them, litigation is not exceptional. It is a cost of operating in this market, and the add-back is removing something real.
The test I use: would I be surprised to see a cost of this kind next year? If the honest answer is no, it belongs in the number regardless of what it is called.
A one-off that recurs annually is not a one-off. It is a category with a different name each year.
Three areas are specific to this kind of company and worth checking directly.
Capitalised development. Engineering salaries moved off the P&L and onto the balance sheet inflate profit today and produce amortisation later. The question is not whether the treatment is permitted, because it usually is. The question is what proportion of engineering is being capitalised, and whether that proportion has drifted upward as the sale approached.
Deferred revenue and the cash it represents. Annual prepayments are wonderful for cash flow and misleading about profit. A business collecting a year upfront has been living on a working capital float. The moment growth flattens, that float stops growing, and cash generation falls even though profit does not move at all. This surprises people badly in year two.
Deferred maintenance on the product. Not an accounting entry, which is exactly why it escapes the analysis. If nothing has been meaningfully rebuilt in six years, there is a cost coming that appears in no historical figure. Treat it as a liability with an uncertain due date, because that is what it is.
Rather than starting with adjusted EBITDA and arguing about deductions, it is more useful to build up from cash.
OWNER EARNINGS, FROM THE BOTTOM UP
Start with actual cash from operations over the last three years, not a single year.
Subtract the capital spending needed to stand still, including the engineering that was capitalised.
Subtract the founder replacement gap: what they were paid, against what replacing them costs.
Adjust for any working capital swing you would not want to repeat.
What remains is the number that services debt and funds distributions. It is almost always lower than the adjusted figure in the deck, and it is the only one you can plan against.
If it is close to the seller's number, that tells you something good about the seller.
This is usually framed as diligence, but the same arithmetic matters if you are not selling anything.
A founder who knows their real owner earnings knows what the business can actually distribute without weakening itself, how much debt it could carry safely, and whether this year's profit reflects performance or a working capital effect that will reverse.
Most founders I meet have a good instinct for their revenue and a much vaguer sense of this figure. That is understandable, since nothing in the monthly accounts presents it, but it means decisions about hiring, distributions, and reinvestment get made against a number that is not quite real.
The exercise takes a day or two, once. It is worth repeating annually.
The deck's number is designed to be defended. This one is designed to be relied upon. Only one of them will still be true in a year.
— Ahmad
THIS WEEK'S MOVE
Write down what it would cost to replace yourself, as a list of roles rather than one salary. Twenty minutes. No conclusions yet.
If you run a profitable software company, start with a free AI roadmap. It takes 60 seconds and shows where you stand and your next three moves.
One operating framework, every Thursday. Free, and you can unsubscribe anytime.